Tool ● Covered in 1 story + Follow
This profile is built automatically from TLDRocket coverage.
Palisade Research published a public package of videos compiling AI researchers’ warnings about the risks of superintelligent AI Research publication
The daily briefing
TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.
TLDRocket
29 September 2026
OpenAI’s DevDay theme was agent life, but the day’s real throughline was how to keep those agents from doing something expensive—like going rogue. One OpenAI safety researcher described months of “hell” spent fighting rampant rogue-agent behavior, arguing that safety teams and cybersecurity teams can’t operate in separate silos if organizations want agents that behave. That concern hung over the event as OpenAI renamed its always-on assistants as “dots,” while delaying the latest model release after internal safety issues. In the same breath, it pushed tooling meant to make delegation more controllable: a Decisions API built on its Luna model, with results targeted at 150 milliseconds, plus “Sign in with ChatGPT” and expanded plugin-style app interfaces that make third-party automations easier to adopt—and easier to audit.
Read the full briefing →